CODEWITHSUNDEEP
pandaspython-3.8
Zanam
Zanam

Reputation: 4807

Pandas converting mixed datatype column to float

I have a dataframe with one column as follows:

df['mixed'].values
array([' GC', '345', '69', '28-'], dtype=object)

I tried the following:

df['mixed'].astype(float, errors='ignore')

But it doesn't do anything. I was expecting the following:

df['mixed'].values
array([' GC', 345, 69, '28-'], dtype=object)

Upvotes: 0

Views: 1280

Answers (2)

Shaido
Shaido

Reputation: 28322

If there is an error during the conversion the original object will be returned. The documentation for astype [emphasis mine]:

errors{‘raise’, ‘ignore’}, default ‘raise’ Control raising of exceptions on invalid data for provided dtype.

  • raise : allow exceptions to be raised
  • ignore : suppress exceptions. On error return original object.

To obtain the expected output, you can instead use to_numeric and setting errors='coerce' which will set any invalid rows to NaN. These can then be set to the original value using fillna (see the answer by BENY).

Upvotes: 1

BENY
BENY

Reputation: 323226

You can do pd.to_numeric

pd.to_numeric(df['mixed'],errors='coerce').fillna(df['mixed']).tolist()
[' GC', 345.0, 69.0, '28-']

Upvotes: 1

Related Questions